徒步导航系统中的自适应步态优化检测方法研究

针对步态检测的可靠性和准确性成为制约个人徒步导航定位系统高精度定位的主要因素，提出一种基于K均值聚类参数调整的自适应步态检测算法。首先研究行走过程的足部运动规律，并分析惯性测量单元输出加速度和角速率的数据特征，提出采用比力幅值滑动方差作为检测依据的步伐检测方案；分析滑动方差窗口参数、初始检测阈值参数和误检修正时间参数对于检测结果的影响并指出三者的关系，确定窗口参数、初始阈值选择的一般准则以及利用K均值聚类自适应确定时间参数以纠正初始检测结果中误检步态的可行性。利用实验室现有MEMS传感器开展了5组不同状态下的徒步实验，实验结果验证了步态检测算法的可行性和有效性。

Research on Adaptive Gait Optimization Detection Method in Pedestrian Navigation System

According to that the highly positioning precision of pedestrian navigation system depend heavily on the re?liability and accuracy of gait detection,an adaptive gait detection method was proposed,which adjust parameter based on K-means clustering. Foot motion during walking phases was studied firstly,and analyzed the data characteristics of inertial measurement output acceleration and angular velocity,then proposed gait detection scheme which using the amplify of specific force as its detection basis. The three key parameters,the window size of sliding variance,the ini?tial detection threshold and the time parameter in false correction and their relationship were also analyzed respective?ly,and find out the basic selection rules of the widow parameter and the initial threshold,while the time parameter can be adaptively selected using K-means clustering in order to correct the false detected gait in initial detection result. Five groups walking experiments under different conditions was carried out using laboratory existing MEMS sensors, the experimental results verified the feasibility and effectiveness of the proposed gait detection algorithm.